Related papers: Training Uncertainty-Aware Classifiers with Conformalized Deep Learning

URL: http://arxiv.org/abs/2205.05878v1
Date: Thu, 12 May 2022 05:08:10 GMT
Title: Training Uncertainty-Aware Classifiers with Conformalized Deep Learning
Authors: Bat-Sheva Einbinder, Yaniv Romano, Matteo Sesia, Yanfei Zhou
Abstract summary: Deep neural networks are powerful tools to detect hidden patterns in data and leverage them to make predictions, but they are not designed to understand uncertainty. We develop a novel training algorithm that can lead to more dependable uncertainty estimates, without sacrificing predictive power.
Score: 7.837881800517111
License: http://creativecommons.org/licenses/by-nc-nd/4.0/
Abstract: Deep neural networks are powerful tools to detect hidden patterns in data and leverage them to make predictions, but they are not designed to understand uncertainty and estimate reliable probabilities. In particular, they tend to be overconfident. We address this problem by developing a novel training algorithm that can lead to more dependable uncertainty estimates, without sacrificing predictive power. The idea is to mitigate overconfidence by minimizing a loss function, inspired by advances in conformal inference, that quantifies model uncertainty by carefully leveraging hold-out data. Experiments with synthetic and real data demonstrate this method leads to smaller conformal prediction sets with higher conditional coverage, after exact calibration with hold-out data, compared to state-of-the-art alternatives.

Err

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.